Spain stopped building. It started squeezing.

Between 2003 and 2008, Spain added 70 homes for every 100 new residents. Between 2015 and 2024 it added 46. The number of households, people actually setting up a home of their own, fell further, from 64 to 37 per 100 new residents. The difference did not disappear. It was absorbed by more people living in each home.

Where the growth came from

Both periods of growth were driven by immigration. In 2003–2008 Spain gained 4.4 million residents, three quarters of them born abroad. In 2015–2024 it gained 2.7 million, and the foreign-born population grew by more than that: the number of residents born in Spain fell by about 0.9 million. This essay is about how the housing stock responded to population growth. That growth has come from abroad, and where immigrants settle can be predicted from where earlier migrants from the same countries lived, which makes it possible to measure the response cleanly.

Three ways to make room

When a place gains residents, its housing market has three ways to make room. It can build. It can turn empty flats and second homes into main residences. Or more people can share each home. Prices are not a fourth way. They are the signal that pushes the market towards one of the three, and how far they must rise depends on how easily the first one works.

In Working Paper 001 I measure the first directly: for every 1% that immigration adds to a province's population, how much does its housing stock grow? The answer changed completely between the two periods.

In 2003–2008, a province whose population grew an extra 1% expanded its housing stock by about 0.7%. Since 2015, the response has been zero: construction stopped following population. Rents did respond, rising about 1% for every 1% of population added, and house prices about four times as much.

This does not contradict Figure 1, where Spain still added 46 homes for every 100 new residents. That figure counts all building in the country. The estimate asks something narrower: did the provinces that received more immigrants build more than those that received fewer? Since 2015 they have not. Homes have gone up, but not where the new residents went.

Where building survived

The 2008 bust left about half a million new homes that nobody bought, very unevenly spread: more than 5% of the housing stock in Castellón, Toledo and Almería, about 1% or less in Bizkaia and Málaga. If that stock had been a reserve, the provinces that held most of it should have filled it first and built later. That is not what happened. The unsold homes were not drawn down faster where more people arrived; most are in places where the new residents did not go. And new building did respond to arrivals, but only in the provinces the bust had left with little unsold housing: there, each extra 1% of population brought new starts equal to about 0.3% of the housing stock. Where the bust left most, it brought almost nothing. The crash appears to have switched off building where it hit hardest, without leaving homes where they were needed.

Why not a figure for each province? I tried. Estimated one province at a time, the numbers are mostly noise: the differences between provinces are smaller than the uncertainty around each one. What the data can show reliably is the national figure for each period, and the provincial traces of what replaced construction.

Fuller homes

The clearest trace of the missing construction is how many people live in each home. For decades Spanish households have been getting smaller, and between 2003 and 2008 the number of residents per household fell in every province but one. Since 2015 it has risen in 15 provinces. Switch between the two periods on the map.

The map uses the Ministry's estimate of households, which is partly modelled. The population register gives a direct count of how many people are registered in each home, and it tells the same story: in the provinces predicted to receive more immigrants, the number of people per home rose by about 0.8% for every 1% added to the population.

The increases are concentrated on the Mediterranean coast and in the islands: residents per household rose by 9% in Girona and by 8% in Santa Cruz de Tenerife and the Balearic Islands. Barcelona added about 7 homes for every 100 new residents, Madrid about 13. In much of the interior, where population is flat or falling, households kept shrinking as before.

The provinces where homes became fuller are the ones whose population grew most through immigration. Before 2015 there was almost no relationship between the two. Since 2015 it is strong: each extra point of population growth from immigration goes with about 1.4% more residents per household, and immigration alone accounts for over half of the differences between provinces.

Who moves in, who moves out

Sharing homes is one way to make room. Another is for people already living in a place to leave. In the model behind the working paper, what residents do depends on a contest between two forces. A growing place pulls people in: jobs, services, the life of a city. Call that pull a. Growth also squeezes those already there, above all through what they pay for housing. Call that squeeze Φ. If the pull wins, a > Φ, and residents move in alongside the newcomers. If the squeeze wins, a < Φ, and residents leave, passing part of the pressure on to somewhere else.

On average, the pull has won. Between 2015 and 2024, the provinces that gained more residents from abroad also gained more Spain-born residents from the rest of the country, about 0.3 for every newcomer from abroad, close to the paper's estimate. The map shows where that breaks down.

Most provinces that grew fast also attracted Spaniards: Málaga, Tarragona, Girona, the islands, Valencia, Alicante, Madrid. There, the pull wins. Barcelona is the clear exception. Immigration added 12% to its population, the most in the country, while it lost 1.2% of its Spain-born residents to other provinces, three times its yearly rate before 2015. Many went next door: the yearly net arrival of Spaniards rose about threefold in Tarragona and fourfold in Girona. In Barcelona the squeeze won, and part of the pressure moved along the coast. Around Madrid, Guadalajara and Toledo also gain residents from the capital, but that overflow is older; it was stronger before 2015 than since.

A province's position on this chart also reflects forces unrelated to population growth. Much of the interior has been losing young people to the cities for decades, which is why Jaén, Badajoz and Córdoba sit at the bottom left. The chart describes what happened in each province; it does not measure each province's pull and squeeze separately.

Is Spain unusual?

A simple question can be asked of any country: do the regions that grow more also add more homes? Using population and dwelling counts from the 2011 and 2021 European censuses, county estimates for the United States (2010–2020) and census divisions for Canada (2016–2021), I measure, within each country, how much faster the housing stock grew in regions whose population grew one point faster. A value of one means housing kept pace with people. A value of zero means growing regions got no more housing than stagnant ones.

Spain has the lowest value in the sample: −0.28 for 2011–2021. The regions that gained population added slightly less housing than those that lost it. In the previous decade the figure was 0.29. Germany, Finland, the United States and Canada sit between about 0.6 and 0.75. Sweden, where Stockholm's housing shortage is well documented, is close to zero. The Spanish figure is not a quirk of the census: the Ministry's own provincial series gives −0.32 for the same decade.

Population growth has different sources across countries, so a closer comparison looks only at growth from immigration. The 2021 European census records, for every region, the residents born abroad who arrived since 2010. Among countries where they reached at least 3% of the population, Germany's regions added about 0.46% more homes for every point of such arrivals, and Denmark's about 0.25; Finland's built more still. Austria, Belgium and Italy show no relationship. Spain and Sweden have negative estimates, though too imprecise to rule out zero. For US counties, using net international migration, the figure is 0.43.

What follows

Sharing is the least visible way to absorb population growth. It does not show up in a price index, and its cost falls on those looking for a home: newcomers, young people leaving their parents, families who need more space. It is also the way that remains when the others are closed. Construction has not responded. Residents, on average, are still moving towards the places that grow, except around Barcelona, where the squeeze has begun to push them out. What is left is sharing.

The question this raises is not how to stop population growth, but how a growing country should make room. A housing market that can build turns new residents into new homes. One that cannot turns them into higher rents and fuller flats.

Those rents are paid by renters, and renters are mostly young and on lower incomes: half of households headed by someone under 30 rent at market prices, against one in twenty of those over 65. The rise in house prices goes the other way, to owners, who are older and richer. When a country stops building where people go, the cost of making room falls on the people looking for a home, and the gain goes to those who already have one.